Adaptive Surface-GKP Code Simulator

Python implementation for Monte Carlo simulation of the surface-GKP code with adaptive stabilizer measurement scheduling and minimum-weight perfect matching (MWPM) decoding.

Project Structure

• main.py: Entry point that parses arguments, runs simulations, and outputs results.
• simulator.py: Core simulation classes implementing different measurement strategies.
  • SurfaceGKPDefaultTrackingSimulator: Standard measurement strategy with all stabilizers measured every round.
  • SurfaceGKPInfoMeasureWeightTrackingSimulator: Adaptive strategy that selects optimal measurement policies per round.
  • SurfaceGKPInfoMeasureWeightTrackingMeasureSimulator: Adaptive strategy with configurable forced measurement/skip rounds.
• simulator_base.py: Base classes and shared functionality for simulators.
• decoder.py: Minimum-weight perfect matching (MWPM) decoder implementation.
• graph.py: Graph structures and utilities for decoding algorithms.
• noise.py: Noise models for GKP codes and circuit operations.
• utils.py: Utility functions for simulation, argument parsing, and result saving.

Dependencies

• Python 3+
• NumPy
• NetworkX (for graph operations)
• Matplotlib (for visualization, if needed)

pip install numpy networkx matplotlib

Usage

Run the Monte Carlo simulation via the command line:

python main.py [options]

Examples

1. Standard surface-GKP code:

python main.py --distance 3 --n_round 3 --mode default_tracking --n_iter 1000

2. Adaptive measurement with distance-5 code:

python main.py --distance 5 --n_round 5 --mode adaptive --n_iter 1000

3. Adaptive with forced measurements on specific rounds:

python main.py --mode adaptive_meas --target_rounds "first,last" --skip_rounds "middle"

Arguments

Argument	Short	Type	Default	Description
--distance	-d	odd int ≥ 1	3	Surface code distance
--n_round	-r	int	3	Number of noisy surface stabilizer measurement rounds
--n_iter	-n	int ≥ 1	1	Number of Monte Carlo iterations
--sigma	-s	float	1.0	Circuit-level noise standard deviation
--sigma_GKP	-g	float	1.0	GKP state preparation noise standard deviation
--sigma_idle	-i	float	= sigma	Idle noise standard deviation
--mode	-m	str	default_tracking	Measurement strategy (see modes below)
--with_info	-w	bool	True	Use syndrome information in MWPM weights
--clean	-c	bool	False	Noiseless GKP inner-code measurements
--workers	—	int	0	Parallel workers (0=auto, 1=serial)
--verbose	—	bool	False	Print detailed per-round information
--log	—	bool	True	Save results to CSV
--csv_path	—	str	results/results.csv	Output CSV file path

Measurement Modes

Mode	Description
default_tracking	Measures all stabilizers every round with noise-tracking MWPM weights
adaptive	Selects optimal measurement policy per round (Z-only, X-only, both, or neither)
adaptive_meas	Adaptive with configurable forced measurement/skip rounds
adaptive_round	Adaptive with forced full measurement every d-th round
skip	Skips all stabilizer measurements except the final round
skip_round	Measures only at distance-multiples and final round

BibTex